The ISP Column
A monthly column on things Internet
December 2007
Geoff Huston
On the Hunt for Critical Internet Resources
I’m writing this column in November, and that means that its time for the travelling circus known as the
Internet Governance Forum (IGF) to come down to earth, unpack its tents and sell tickets for its annual song
and dance routine. The script for this year’s show has been changed, and after being excluded from the main
arena last year at the Athens gig, the headline act of “Critical Internet Resources” is taking a starring role this
year in Rio. Some folk are even saying that it’s the single most contentious issue to be scheduled at this year’s
IGF show.
So what are “Critical Internet Resources” anyway? If folk are going to spend all this time, energy and carbon
emissions travelling to Rio to talk on this topic, then wouldn’t it be helpful to understand what it means in the
first place?
There are probably a number of ways to answer this question, so in this heavily opinionated column I’d like to
look at the range of possible answers to this question.
The first answer lies in looking at the question from a purely personal perspective. I’m sitting here typing this
on my laptop and the most critical resource for me right now is the reserves of battery power for my laptop.
Without that then I’m hosed, and everything I’ve done in the past hour or two would also be lost, which
makes battery power a pretty critical resource for me at the moment! But that’s more about my laptop and
not the Internet per se. What’s critical for me with respect to the internet right now? Right now I’m connected
to the network using a WiFi connection, so without this connection the internet ceases to exist for me. So it’s a
case of no WiFi connection then no Internet either, so maybe that’s my critical internet resource. So with both
of those resources at hand, then I’m set. So where I sit right now my needs in the critical internet resource
department appear to be decent batteries and a good WiFi connection. Of course here I’m talking a pretty high
threshold interpretation of “critical” here, in that “critical” means that if its not there then neither is the
Internet. This implies that “critical” falls into the highest category of resources, namely those resources that if
they are not available then the Internet is not available. So I’m still nominating my battery power and my WiFi
connection as my most critical internet resources. But somehow I don’t think that the IGF spaceship has come
down to earth for a week just to recharge my laptop’s batteries and install a few more WiFi base stations, so I
guess I must be missing the point here.
So maybe it’s a case of asking others by assembling a number of Internet users’ views and seeing what comes
out as being the most popular in terms of critical internet resources. So maybe I should conduct a critical
internet resource survey. Fortunately, Paul Wilson has already conducted such a survey in recent weeks, so I’ll
save myself the effort and just look at what he found out when he asked some Internet users what they
thought fell into the category of “Critical Internet Resources” The results of his survey are shown in the
following table.
What’s clear at the outset in scanning through the results of this informal survey is that the users Paul
surveyed are clearly aware of the many diverse inputs that make up the Internet. Its also evident that the
perception of what’s “critical” to the Internet is certainly quite a broad collection of inputs, and some of them
are a little surprising. I can see the logic in nominating IP addresses, names, and standards as being critical to
the Internet, but I’m personally not yet convinced that Internet Exchange Points are in the same category as
being “critical” to the Internet. Sure IXP’s can produce locally advantaged outcomes by making local resources
and local communications faster and cheaper for local users, and also encourage continued diversity and
competition in the access market, but I’m not sure that its gets on to the “critical” list as in the internet
absolutely relies on it. On the other hand cost is a real consideration for many users and the Internet is only
available if it can be made available within a certain price threshold. So anything that makes the Internet
affordable to users is properly considered to be a “critical” factor here. Infrastructure elements of the physical
network appear to be commonly thought of as critical resources, although I recall that some element of the
rationale for a stateless packet-based network architecture was to remove the critical role of any single
physical infrastructure element and instead provide the network with the ability to route around various forms
of component failure. Even so, this is an interesting list, in that it illustrates the diversity of inputs that support
the operation and use of network, and in their own way each of these components is critical.
Category
Administration
DNS
ICANN/IANA
Standards
Names/Numbers/ASNs
IP addresses/RIRs
Administration Total
Applications
Email
other
Search
Communications
Information
WWW
Commerce
25
10
22
11
7
9
12
96
Electricity
funding
Openness
Human Resources
2
2
8
3
15
Connectivity
Core
Devices
International
IXP
Lastmile
Peering
QoS
Security
Ubiquity
Routing
10
13
3
5
3
9
4
3
9
3
16
78
Applications Total
Environment
Environment Total
Infrastructure
Infrastructure Total
Grand Total
Total
30
3
5
6
10
54
243
What I find of most interest here is the recognition of the Internet as the sum of its application environment.
There is no doubt that one of the more revolutionary aspects of the Internet has been in information models,
and the advent of search engines has indeed been truly revolutionary. It seems so natural these days to type
in anything at all into a search engine and generally get decent results back. Indeed its those times when the
outcomes of the search engine don’t appear to be logically associated with the question that we regard as
strange. And its only been a decade since we had nothing at all in this space. Equally, we now regard IM as
staple diet of messaging and for many it is the critical application of the Internet. So, increasingly, we now
are starting to describe the Internet in terms of the services and user environment that it enables, rather than
the infrastructure requirements that are necessary to provide it, and now starting to look at this use
environment as the internet’s most critical aspect. So maybe the hunt for critical internet resources should be
Page 2
up at the application level, and looking at information and context as being the critical resources and all else
as being derivatives from these basic and critical capabilities.
But in that rarefied atmosphere of the heights of the international Internet policy and governance debate the
term “Critical Internet Resources” appears to have a very different meaning. This phrase appears to be a code
phrase that is intended to describe only the administration of the Internet’s naming and addressing domains.
Now if you take a liberty here and broaden this definition out a bit and include DNS names and IP addresses
themselves as well as their administration, then a case could be made that these are indeed critical to the
Internet. Without addresses then what goes into a packet header? How are packets forwarded across the
network? So from that perspective addresses appear to be somewhat critical to the Internet. But what sort of
addresses are we talking about here? Globally unique addresses or private addresses? It certainly appears
that far more of the Internet is numbered from private use address space than globally unique address space.
So maybe its these private use addresses that are the critical Internet resource here, in which case it make
little sense to attempt to connect the dots and infer that the administration of these addresses is a critical
function, given that the addresses are not administered in any coordinated fashion. But even that claim needs
to be put into some perspective. Is my street address a “Critical Postal Resource”? In some countries a street
address is a set of directions that appears to make sense only in a very local context. It appears to be the
same with IP addresses, in so far that they also have a scope and they are important only within that defined
scope.
Well what about DNS names then? The domain name business is an impressive achievement, if only by virtue
of its somewhat surprising size and relative wealth. While the Internet’s carriage and access sectors appear to
be still coming to terms with the harsh brutality of the term “competitive commodity utility” the domain name
registry industry is apparently still trying to figure out what to do with all that money! From domain name
tasting to sitefinders to IDNs the name business is apparently big business and an attractive one at that. But
are names a “critical” resource? Well there is no particular shortage of names, and my having a domain name
does not displace your having a name as well. And the telephone system has shown us how to construct a
network whose presentation format is entirely based on digits, so its not immediately apparent for me to see
than names are necessarily the “critical” internet resource.
This hunt for “Critical Internet Resources” is still not coming up with an obvious short list. Maybe its routing
slots. After all, we are told that the routing system cannot grow forever. It has finite capacity, and if routing
does not work, or your address is not described in the routing system, then the Internet simply stops working
for you. But does that make it a “critical” internet resource? There is no single routing system. There are a
collection of local entities that operate their own local routing system and exchange information, and packets,
with other entities, based on the operation of markets in peering and transit. The outcome is important, to be
sure, but I’m still not convinced that this is the “critical” resource for the Internet. Routing technologies come
and go. RIP is one of the ones that’s gone. Other evolve, or will get replaced in time. So its hard to construct
the case that we are stuck with a particular routing technology with particular properties and particular critical
vulnerabilities.
Maybe this hunt is actually a pointless quest. Like many complex engineered systems in today’s world, the
Internet is constructed using a very large collection of supply chains with a large and diverse collections of
players and stakeholders and a huge set of interdependencies. One could make the case that critical internet
resources encompass areas of silicon design and fabrication, power generation, laser technology and fibre
cable manufacture, and so on. Indeed it is possible to make that case that almost every other industry has
inputs into the Internet. And the Internet provides inputs to almost every other industry these days. We’ve
constructed these systems building upon the outputs of other systems, and plugging their outputs into other
systems. Its not a hierarchy, it’s a dense interconnection, and from this perspective the entire fabric is the
“critical resource”. So it really doesn’t make a lot of sense to label one artefact or aspect of the Internet as a
“critical resource” and not another.
What then can we make of the hunt for these “Critical Internet Resources” that is being conducted by some
countries in the hallways and performance arenas of the IGF? Perhaps its just another round in the endless cut
and thrust of international politics, and the real list of “Critical Internet Resources” are those where some
nations feel that other nations have a privileged or unfair position of control or influence. This appears to be a
consistent theme of the World Summit on the Information Society meetings in 2005, and the first IGF meeting
in 2006.
Its true to say that the Internet was in many ways another example of the Law of Unintended Consequences
in action, and the outcomes we have today were never planned in any particular fashion. It was just a
research program that stumbled upon an extraordinarily effective marriage of computers and communications
that then just grew and grew. And the original administrative arrangements for names and addresses were
simply the outcomes of pragmatic decisions to achieve the best results within the available funds. When the
Page 3
entire name and address administrative arrangements become the victim of the Internet’s runaway success
then the successive refinements represented a compromise between what had happened in the past and what
needed to happen at the time.
If you take the view that the oversight of the governance of the name and address distribution infrastructure
is a responsibility undertaken by ICANN by virtue of an agreement between this corporation and the
government of the United States of America, then, yes, there is a nation that appears to be placed in a
privileged position as compared to other nations, and when compared to other arrangements that rely on the
underlying instrument of some form of an international treaty then, yes, this arrangement is different in many
respects. Like many bold innovative experiments in international coordination and the establishment of new
world orders, ICANN stands a strong risk of falling foul of an inherent conservatism in international politics,
where the careful balancing of national interests is seen as being far more critical an objective than any actual
outcomes that may be achieved from the process. From this perspective ICANN is critically reliant on its
acceptance by all players of its legitimacy to operate in this space, and also critically reliant on acceptance of
the proposition that these issues are best addressed in open forums of debate. This is a difficult task, and the
set of outcomes that ICANN can point to as being products of this process do not install an absolute level of
confidence that this process is stable, scaleable, and sustaining. Right now the proposition is not that ICANN
represents an outstanding set of achievements, but that the track record of the alternatives have failed in
various ways in the past and nothing has changed to prevent similar flawed decisions in the future.
But this is a long way from “Critical Internet Resources”. Perhaps this particular matter was never about this
topic in the first place. Perhaps the real matter was a perception that the US has overreached itself in securing
a privileged role in the Internet administrative framework, and this has opened up an avenue for trenchant
criticism in UN related forums. Its certainly the case in the context of the IGF that a number of other nations
have been attempting to make some political mileage out of this situation by labelling these name and address
administrative arrangements with the use of a particular phrase that has extensive connotations into areas of
criticality and crisis, connotations into resources and resource management and connotations with the high
frontier of technology and an increasingly technology-dominated future. What better than to label these
arrangements “Critical Internet Resources?” So when we look at the use of this term in the context of the IGF
we find, not surprisingly, an entire world of politics and related considerations of national positioning in an
international context, but no actual “Critical Internet Resources”! This is, of course, shocking news.
Its probable that you are not going to find the Internet’s most critical resource lurking in your computer or
hidden in an equipment rack somewhere or sitting in a configuration state in the network. You are not going to
find it in an application or service, and you are definitely not going to find the Internet’s most critical resources
lurking in the corridors of the IGF meetings. I suspect that, as with any communications network, the most
critical resource of the Internet, and the entire point of all of these public communications networks in the first
place, is, simply, just users like you and me and our driving need to communicate.
Disclaimer
The above views do not necessarily represent the views or positions of the Asia Pacific Network Information
Centre.
About the Author
GEOFF HUSTON holds a B.Sc. and a M.Sc. from the Australian National University. He has been closely
involved with the development of the Internet for many years, particularly within Australia, where he was
responsible for the initial build of the Internet within the Australian academic and research sector. He is author
of a number of Internet-related books, and is currently the Chief Scientist at APNIC, the Regional Internet
Registry serving the Asia Pacific region. He was a member of the Internet Architecture Board from 1999 until
2005, and served on the Board of the Internet Society from 1992 until 2001.
http://www.potaroo.net
Page 4